Spatial distribution of soil nutrient content for sustainable rice agriculture using geographic information system and Naïve Bayes classifier

Abstract

This study aims to assist farmers in monitoring soil nutrients, especially phosphorus. To measure the phosphorus content of paddy soil, the TCS3200 converter, as an intelligent sensor, was applied. The geographical information system (GIS) was also involved in this research to map the phosphorus content. In addition, the Naïve Bayes method was applied to classify lowland soil phosphorus status. The result of this study indicated that the Naïve Bayes algorithm could classify lowland soil phosphorus status with a probability of 0.34 for moderate phosphorus conditions and 0.66 for high phosphorus conditions. The sample testing results showed that the error rate was 3% and the success rate was 97%. Testing with a phosphorus-measuring instrument can be carried out by mapping the soil phosphorus status with the ArcGIS software, whereby seven points of medium-phosphorus-status paddy soil and 13 locations of high-phosphorus-status soil samples were determined. This research thus successfully mapped the soil phosphorus.